US 7042573 B2
In order to characterize the optical characteristics of a device, a source of light having a variable frequency with a polarization state which varies linearly with frequency is provided as an input to the device under test. The input light is also passed through a known reference path and is added to the light output from the device under test in a beam combiner. The combined light for the frequencies of interest is split into two orthogonal polarizations which are then detected in a spectral acquisition apparatus and supplied to a microprocessor. The spectral measurements are digitized and curve-fitted to provide optical power versus optical frequency curves. Fourier transforms of each of the curves are calculated by the microprocessor. From the Fourier transforms, the four arrays of constants are calculated for the Jones matrix characterizing the device under test.
1. An optical vector analyzer for analyzing the characteristics of a fiber-optic device under test (DUT) over a frequency range of interest, comprising:
(a) a source of light for providing a varying optical frequency light over said frequency range of interest, said light having an initial polarization;
(b) a spinner coupled to receive said light including:
(1) a splitter for splitting said light from said source into two portions of light;
(2) a first path and a second path including a switch, each path being supplied with one portion of said light;
(3) a spinner combiner for combining light from said two paths into a spinner light output coupled as an input to said DUT and a reference path;
(c) a beam combiner for adding light from said reference path and from said DUT for a plurality of frequencies in said frequency range of interest and for providing an output indicative of optical power, in two orthogonal polarization states, of said combined light;
(d) a spectral acquisition unit for measuring said optical power with respect to wavelength for said plurality of frequencies without the DUT coupled to receive the spinner light output, first with the switch open to obtain first measurement data and then with the switch closed to obtain second measurement data; and
(e) a microprocessor responsive to said measured power and programmed to use the first and second measurement data in calibrating the optical vector analyzer.
2. The optical vector analyzer in
(1) digitize said measured power for said frequencies,
(2) derive respective curves from said digitized power measurements,
(3) calculate the Fourier transform of said respective curves, and
(4) derive from said Fourier transforms one or more arrays of entries for a Jones matrix, thereby characterizing said DUT.
3. The optical vector analyzer in
4. The optical vector analyzer in
5. The optical vector analyzer in
6. An optical vector analyzer according to
7. An optical vector analyzer according to
8. An optical vector analyzer according to
9. An optical vector analyzer according to
10. An optical vector analyzer for analyzing the characteristics of a fiber-optic device under test (DUT) over a frequency range of interest, comprising:
(a) a source of light having an optical frequency content over said frequency range of interest, said source of light providing two mutually coherent light beams, each of said light beams feeding a spinner for providing a spinner light output having a polarization state which varies linearly with said frequency, and said spinner light outputs having different rates of change of polarization with respect to change of frequency of the light beam, one of said spinner light outputs comprises an input to said DUT and the other of said spinner light outputs comprises an input to a reference path;
(b) a beam combiner for combining light output from said reference path and light output from said DUT over said frequency range of interest and providing a detected output; and
(c) a microprocessor, responsive to said detected output and programmed to
(1) digitize said detected output over a plurality of frequencies at said frequency range of interest,
(2) derive respective curves from said digitized outputs,
(3) calculate the Fourier transform of said respective curves, and
(4) derive from said Fourier transforms one or more arrays of constants for a Jones matrix, thereby characterizing said device under test,
wherein all of the optical components in the optical vector analyzer are constructed from optical fiber.
This is a continuation of U.S. patent application Ser. No. 10/005,819, filed on Dec. 14, 2001, the entire contents of which are incorporated by reference.
This application claims priority from provisional application Ser. No. 60/255,077 filed Dec. 14, 2000 and now abandoned.
1. Field of Application
The present invention generally relates to the measurement of the optical characteristics of optical waveguide devices and, in particular, relates to devices for characterizing the optical effect of various devices upon an optical system.
2. Discussion of Prior Art
Waveguide devices such as Bragg gratings, interleavers, couplers, isolators, etc. require accurate characterization if they are to be used in deployable optical fiber networks. Generally this characterization has been accomplished using a few simple specifications such as insertion loss, bandwidth, polarization dependent loss, etc. The devices actually have responses that are significantly more complicated that these specifications can describe. Although the simplification of the characterization is necessary to allow for the use of interchangeable parts, and rational specifications for manufacturers to meet, obtaining a full and complete measurement of the device from which any desired specification can be calculated would have benefits to both manufacturers and their customers.
First of all, full device characterization would permit computer system models that are much more accurate, and would inherently contain aspects of the device that have not been explicitly specified. Second, in cases where manufacturers use different specifications, or consumers require a new set of specifications, devices could be accurately compared, or new specifications developed from the complete measurement function stored in a database. For most devices this measurement function would consist of four complex functions that would occupy less than 200 kB of storage. Finally, if the complete characterization of the device can be achieved with a single instrument, then final testing of devices would be greatly simplified while remaining completely general.
It is worthwhile to describe some of the fundamental differences between the characteristics of fiber-optic links, and fiber-optic components (excluding fiber). Fiber-optic links are generally very long (>50 km), and have very broad spectral features that vary on the scale of tens of nanometers. Because of the great lengths involved and high launch powers used to over come the loss, nonlinear interactions of light in the fiber are of significant interests. These nonlinear properties are not easily reduced to a simple transfer function matrix. In very long fiber-optic links such a submarine links, optical amplifiers are used to regenerate the signal and overcome loss. These amplifiers often operate in saturation which is another non-linear effect not captured in a simple transfer function matrix. Generally, fiber-optic links are not well characterized by simple transfer functions.
The relatively short path lengths (<1 m) within fiber optic components prevents any significant non-linear behavior. Because many passive fiber-optic components are filters used to separate channels closely spaced in wavelength, they are designed to have rapid variations in their properties, as a function of wavelength. Optical filters often display variations on the scale of tens of picometers instead of tens of nanometers as seen in fiber optic links. Not only does the transmission amplitude of these functions vary rapidly, but the phase response does as well, and so, in high bit-rate applications, the phase response must be accurately measured as well. Also, due to the fabrication techniques, many filters display some variation as a function of polarization. Again, in high bit-rate systems, this variation must be accurately characterized. It is in these last two measurements, optical phase, and polarization dependence, that the inventions presented here excel.
The underlying technology used was initially developed by Glombitza and Brinkmeyer, and published in 1993 (“Coherent Frequency-Domain Reflectometry for Characterization of Single-Mode Integrated-Optical Waveguides,” Journal of Lightwave Technology, Vol. 11, No. 8, August 1993). No patents were filed by either author with regard to this technology. Patents of a similar nature are U.S. Pat. No. 5,082,368 “Heterodyne optical time domain reflectometer,” which employed an acousto-optic modulator, and U.S. Pat. No. 4,674,872 “Coherent reflectometer,” which employed an optical phase shifter, and has expired due to nonpayment of maintenance fees. A co-pending U.S. patent application Ser. No. 09/606,120 filed Jun. 16, 2000 by Froggatt and Erdogan for an invention entitled “Single Laser Sweep Full S-Parameter Characterization of Fiber Bragg Gratings,” herein incorporated by reference, expands upon the published and unpatented work by Glombitza and Brinkmeyer to describe how the measurement of the phase and amplitude of the time domain response of a system can be used to determine the frequency domain (spectral) response of the system as well via an inverse Fourier transform. U.S. Pat. No. 5,896,193 for “Apparatus for testing an optical component” also describes an interferometer much like that described by Glombitza and Brinkmeyer although the patent filing on Feb. 14, 1997 post dates the Glombitza and Brinkmeyer article publication date of August 1993 by over three years.
The Jones Matrix Representation
An optical fiber component supporting two polarization modes can be fully described using a wavelength dependent Jones matrix. This matrix describes the transfer of energy from the device input to the device output, encompassing the full effects of polarization dependence. The Jones Matrix relates the input electric field to the output electric field by:
This four element two-by-two matrix is completely general and includes all aspects of the device behavior. If the four complex numbers (a, b, c & d) can be measured as a function of frequency, then the device has been completely characterized. It is therefore an object of this invention to measure these four quantities, and from this basic physical measurement derive any and all of the desired parameters describing the device.
It is an object of the present invention to supply a controlled light input to a fiber optic device and then compare the output from the device with the input to characterize the effect of the device on such light.
In the broadest sense, the present invention involves the supplying of light having an optical frequency content over the frequency range of interest where the light has a polarization state which varies linearly with the frequency. This light is provided to the device under test. The light is affected by the device and is output therefrom. The affected output light is combined in a combiner with the input light over the frequency range of interest. The combined light from the combiner is detected and digitized over a plurality of measurements in the frequency range. A microprocessor derives power curves from the digitized combiner output. The microprocessor calculates the Fourier transform of the respective curves and derives from the Fourier transforms four arrays of constants for the Jones matrix characterizing the device.
In a specific embodiment, a source of variable frequency light which could be a tunable frequency laser is provided where the light has a similar polarization. The light is supplied to a “spinner” which provides output light having a polarization which varies linearly with the changing frequency of the light. The light from the spinner is supplied as an input to the device under test and also to a beam combiner. The resultant light from the device under test is also supplied to the input of the combiner and its output is supplied to a polarization beam splitter. After having the combined light split into two orthogonal polarization states, the two orthogonally polarized beams of light are supplied to respective detectors. The outputs from the detectors is digitized over a plurality of measurements taken at differing frequencies and respective curves power are derived. A programmed microprocessor calculates the Fourier transform for each of the curves and derives four arrays of constants for a Jones matrix characterizing the device under test.
In order that the present invention may be more clearly understood, examples of various embodiments thereof may be described by reference to the following drawings, in which:
For simplicity of understanding, similar components will be similarly numbered throughout the several views.
The inventions described here can readily measure the matrix above, however, the basis vectors used to describe the electric field is left unknown. This topic, and the reason that it can be neglected is described in the section “Unitary Translation and Polarization Characteristics”.
Unitary Translation and Polarization Characteristics
Telecommunications systems use optical fiber that supports two polarization modes. Some effort has been expended to develop single-polarization mode optical fiber, but the losses remained too high to permit their use in long-distance transmission systems, and so, two polarization modes are supported in all currently installed fiber-optic links, and will continue to be supported in all links installed in the foreseeable future.
A “supported” mode is one that will propagate over long distances. The two modes supported in telecom links can be described as “degenerate” meaning that light in each mode propagates down the fiber at nearly the same speed. The difference in the speed of propagation between the modes is referred to as polarization-mode-dispersion. For short lengths of optical fiber (<100 m), this difference in propagation velocity is so small that the delay difference is essentially zero. Over very long transmission distances (>50 km), the very small difference in propagation speed can cause significant differences in the delay between polarization modes. This accumulated difference now forms the barrier to increased data rates in optical fiber. As a result, methods to compensate for polarization mode dispersion are now at the center of many optical fiber communications research programs.
When the polarization-mode-dispersion in an optical fiber is kept small, the two modes are considered to be degenerate. A consequence of degeneracy is that energy can very easily move between the supported modes in the fiber. This freedom of movement between the supported modes in the optical fiber means that the polarization state in optical fiber is essentially uncontrolled. Mild bends, external stresses, and thermal changes can all change the state of polarization of light in optical fiber.
Although the input and output polarization states for a segment of fiber can be considered to be independent, the lossless nature of fiber transmission does impose some restrictions on the output state of the light. First of all, the amplitude of the output light must be the same as the input light (which is essentially the definition of lossless). Less obvious, but of great importance is the maintenance of orthogonality, i.e., when the dot product of two fields is zero.
Orthogonal polarization states do not interfere with one another, and by combining the states in different phases and amplitudes, we can represent all states of polarization. Pairs of orthogonal states, therefore, represent an important physical construct. In a lossless system, such as an optical fiber, orthogonal inputs result in orthogonal outputs. Any section of optical fiber can then be modeled as a “unitary translation.”
What does the concept of unitary translation mean to optical fiber systems? A good analogy is that of a three dimensional object in space that is rotated. Take for instance, a cat. If I have a cat and rotate it by some amount left to right, and then another amount front to back, potentially every point on the cat changes location, however, all of the important properties about the cat remain the same. Equivalently, if an arbitrary section of optical fiber is placed between an instrument and a device, the measurement of the device is rotated by two unknown angles, but the properties of the device are all measured correctly. And because polarization states are not controlled in fiber-optic communications systems, the end user of the device cannot be certain of the device's “orientation” (with respect to the incident polarization) in the system, and must be assured that the device will work for all orientations.
The concept of unitary translation can be extended further to measurement error. Any measurement error that can be represented as a multiplication by unitary matrices then becomes irrelevant to the measurement. The ability to fold such measurement errors, or unknowns, into the unknown fiber lead is what allows the present invention to be constructed in a tractable and economical way.
The spinner 16 causes the state of polarization of the light to rotate through a great circle around the Poincaré sphere as function of wavelength. The relationship between the angular change of the polarization state thought the great circle and the frequency of the light must be linear. The period of the variation (i.e. the frequency change needed for 2 pi rotation) must be greater than the spectral resolution of the acquisition system. In the case of a tunable laser source, this is the instantaneous linewidth of the source, and, in the case of an OSA (Optical Spectrum Analyzer, a widely used commercially available device), this is the resolution bandwidth of the OSA.
The output from source 12 is provided to a reference path 18 and to the input of the DUT 10. The reference path is a method of transmitting some portion of the light from the source 12 to the beam combiner 20. This path must have a Jones Matrix with a determinant that is not close to zero. In other words, this path cannot be a polarizer, and, is preferably a lossless optical transmission such as a fiber, but it need not be perfectly or substantially lossless. The matrix of the reference path must be invertible in a practical sense. The transmission path to the DUT need only satisfy the same criteria as the reference path matrix. The transmission path from the DUT should be substantially lossless, or lossless to the degree of the accuracy required by the measurement.
The output of the DUT 10 along with the output of the reference path 18 is supplied to the input of a beam combiner 20 (indicated by dotted line box). The combiner adds the light output from the DUT 10 and the reference path 18 for a plurality of frequencies in the frequency range of interest and splits the sum into two orthogonal polarizations. In preferred embodiments, the beam combiner 20 can be a coupling structure 22, such as a bulk-optic beam splitter or a fused tapered coupler, which supplies a combined light signal to a polarizer 24 which breaks the field into two orthogonal polarization states.
The two orthogonal polarizations from beam combiner 20 are supplied to the microprocessor 26 (indicated by dotted line box). In a preferred embodiment, the microprocessor 26 includes a spectral acquisition device 28 and a mathematical processor 30. The function of the spectral acquisition device 28 is to measure the power as a function of wavelength of the two signals from the polarizer. These signals may be acquired sequentially or in parallel depending upon the application and embodiment.
The spectral acquisition device 28, if the optical source is a narrow linewidth tunable source such as a laser, will be a pair of detectors from which the power is sampled as a function of the frequency of the tunable source. Also, if the source is a narrow linewidth tunable source, then it may be necessary to send a small part of the light from the source to the spectral acquisition unit so that the wavelength may be accurately measured (as shown by the dotted line 13 connecting source 12 and microprocessor 26 in
The mathematical processing unit 30 includes a digitizer to convert the spectral measurements made in device 26 to digital form if necessary and is programmed to derive curves fitting the spectral outputs, where the curves are optical power versus optical frequency curves. Fourier transforms of each of the curves are calculated by the processor. From the Fourier transforms, four arrays of constants for a Jones matrix characterizing the DUT are calculated. Knowing the four arrays of constants of a Jones Matrix for the DUT allows total knowledge of how and in what fashion any device under test will affect an input signal and thus affect a fiber optic system.
Details of the “spinner” 16 are shown in
The spinner 16 puts the light from one path into the x-polarization, and light from the other path into the orthogonal y-polarization. Because the path lengths of the two components are unequal, the relative phase between the x-polarized light and the y-polarized light is strongly wavelength dependent. As a result the polarization state of the light exiting the polarization beam combiner 36 varies from linear to elliptical to right hand circular to elliptical, etc. as shown in the 9 graphs of
If the polarization state is viewed on the Poincaré sphere as shown in
In order to monitor the source of light, in a preferred embodiment some of the light exiting the polarization beam combiner 36 of the spinner 16 is picked off by the single-mode coupler 38 as shown in
While the detectors 56 & 58 are shown outside of the dotted line encompassing the polarizer 24, they could be included within the polarizer and their output signal provided to the spectral acquisition block 28 as shown in
The polarization controller 52 can be adjusted to one of two possible alignments. The adjustment should be made with no light coming through the measurement path. In one polarization controller alignment—referred to as the “nulled alignment”—the two linear polarization states are mapped to the two linear polarization states of the polarization-beam-splitter 54 when they pass through the reference path. In this alignment, the relative phase of the two states of polarization at the polarization-beam-coupler 43 does not affect the power splitting ratio at the polarization-beam-splitter 54 for reference path light, and as a result, no fringes are observed when the laser is tuned.
In the other polarization controller alignment—referred to as the “maximum contrast alignment”—the great circle circumscribed by the reference light is made to intersect with the two linear polarization states of the beam splitter 54. Whenever the reference polarization state passes through one of the linear states of the beam splitter, all of the reference power is incident on one detector, and none is incident on the other detector. The reference power then oscillates between the s and p detectors 56 & 58, respectively, going to zero on one detector when the power is maximized on the other detector. This output in a preferred embodiment is provided to the spectral acquisition unit 28.
If we look at the Fourier transform of the fringes, we get a time domain response, with the time domain response of each Jones matrix element arriving at a different time. By properly selecting the complex valued points associated with each matrix element, we can then use an inverse Fourier transform to compute the frequency domain (spectral) Jones matrix elements. While each term is multiplied by a phase error, the result of this phase error is equivalent to an arbitrary section of lossless optical fiber added to the device, and thus has no effect on the measured properties of the device.
We can begin by representing the DUT or “device” as a simple diagonal 2×2 matrix with complex, frequency-dependent entries,
In order to measure the complete response of the DUT, it is necessary that two orthogonal polarizations be incident upon the device. These states cannot be incident simultaneously, because the linear superposition of these states will form a new, single polarization. In order to perform a rapid measurement that will completely characterize the device, it is desirable to use a polarization state that changes rapidly with frequency, such that the device response does not appreciably change for one complete cycle through polarization states. An optical device to produce this rapidly varying polarization using a PM splitter and a polarization spinner 36, has previously been disclosed in
In the “spinner” of
The normalized field emerging from this device is then given by,
In order to meet the criteria above that the polarization state vary rapidly enough that the response of the DUT does not change appreciably through a single cycle, the delay, τp, in the network above must be greater than the impulse response length of the device, τh. The fiber network that transfers this field from the 50/50 beam splitter 41 to the DUT will have largely uncontrollable polarization properties. It is desirable to impose two restrictions on this network. First, it have no Polarization Dependent Loss (PDL), and second, the matrix describing the network be constant as a function of frequency over the range of frequencies that is of interest. Of these two requirements, the zero PDL criteria is the most difficult to satisfy. The matrix for such a network is,
Here, we will digress a moment to discuss the parameter, θ, and what might be done about it.
As shown in
By adjusting the polarization controller 52 to achieve full contrast, θ is set to
Examining these detected signals it can be seen that the elements of the matrices have been shifted to separate locations in frequency space, with a phase shift. The relative phases between the elements in a Jones matrix are important and must be known to some degree as discussed later. So, by selecting terms based upon their location in frequency space, and, also making a best guess at the polarization spinner delay, τp we can extract the following field responses,
Since we maximized the fringes that oscillate at a frequency of ωτp, we can easily determine τp for each scan, thus allowing for variations in the spinner path imbalance from scan to scan due to temperature. With the ωτp term removed from the exponents, and any static phase error lumped into ψ and φ, we can then construct the matrix,
Note that the measured matrix can be obtained from the original matrix by multiplication with two unitary matrices. This is not a transformation of basis, because the two matrices are not inverses of one another. As a result, diagonalizing the measured matrix does not recover the original matrix. All of the important properties of the matrix, Polarization Mode Dispersion, Polarization Dependent Loss, Spectral Amplitude and Group Delay are all maintained. All that has been altered is the polarization state at which a given minimum or maximum will occur. Since polarization in optical fiber is rarely well controlled, this loss of information regarding the device is not detrimental to the vast majority of applications.
In the following embodiments, the mathematics describing the device performance is not substantially altered, although the elements used to implement the measurement are altered. These alterations involve the removal of elements from claims and are not obvious.
No Polarization-Maintaining Fiber Embodiment
The embodiment described above in
At the polarization beam combiner 20, the great circle on the Poincaré sphere through which the polarization state rotates is well defined. Once the light has propagated through a few meters of optical fiber, however, the fiber will have transferred the polarization states to a new and completely arbitrary great circle. As a result, the well defined states produced by the PM fiber system serve no useful purpose to the measurement. We are then free to choose any set of orthogonal states to send through the two different length paths.
As disclosed in U.S. Pat. No. 4,389,090 “Fiber Optic Polarization Controller,” describes a simple apparatus that can be used to map any input polarization state to any other polarization state with little or no loss and little wavelength dependence. The polarization controller consists of three fiber loops that can be tilted independently, and is generally referred to as a set of LeFevre Loops after the inventor.
If we include a polarization controller of any kind (such as the LeFevre Loops discussed above) in a Mach-Zender Interferometer, we can adjust the controller such that the two states, when recombined at the coupler, are orthogonal to one another. This state of orthogonality is achieved by adjusting the polarization controller until one output of the coupler shows no fringes as the laser is tuned. The lack of interference fringes indicates that the two states of polarization entering the coupler through different paths have no common components, and are, therefore, orthogonal. With the creation of a polarization state that rotates rapidly through a great circle on the Poincaré sphere as a function of wavelength, we have achieved an equivalent function to the PM fiber/polarization-beam-combiner device described above. The remainder of the system then operates exactly as the previously described embodiment.
Calibration of a single Spinner Optical Vector Analyzer
In order to describe the calibration procedure, it will be necessary to reference specific points of the optical network within one embodiment of the optical vector analyzer. A schematic diagram of the relevant portion of the network appears in
Before performing any measurements, and after aligning the optics, two measurement scans are conducted with no DUT attached. The first scan is done with the spinner switch open, and the DC signals on the detectors at points w and x are recorded. These DC signals are equal to the optical power incident on each detector multiplied by the gains of the detectors and their supporting electronics, and are frequency-dependent. We will denote them as
In the above expressions ωis the optical frequency and τ is the difference between the times for an optical signal to propagate from point b to point e and from point c to point d. By taking the Fourier transform of these signals, the AC and DC terms can be separated, and then inverse transforms will bring the information back into the frequency domain. We obtain the following information from this scan:
Once the above information has been obtained, it will be used to divide all subsequent measurements in the following way. If the raw Optical Vector Analyzer Jones matrix measurement yields the matrix
During a measurement scan, the detector 62 at point h acquires data at the same time as the detectors 58 and 56 at points w and x, respectively. This data has the same form as Eqs. 42 and 43, and the AC component can be extracted in the same way. We will call this result Sh ac(ω), and multiply the Jones matrix by the phase of Sh ac(ω) as follows:
The next step in calibration is to perform a measurement using a fiber patchcord as the DUT. The resulting matrix, upon being divided by the calibration quantities as in Eqns. 48–51, is stored in memory and is called the reference matrix. It will be denoted as
All Coupler/Polarization Controller Embodiment
The polarization beam-splitter 54 remaining in the design in the description above not only adds cost, but places very tight restrictions on the coupler tolerances in the spinner interferometer. A problem with the above embodiment is that a polarization controller 52 and polarizing beam-splitter 54 would be required for each additional receiving channel if multiple detection channels were required.
These difficulties can be resolved using a double spinner embodiment without polarization control as shown in
The outputs of spinners 16 a and 16 b are applied to the DUT 10, and the reference path 18, respectively. The two outputs are connected to the coupler 22 with the output detected and applied to the microprocessor 26 as before. However, because of the difference in polarization rate of change, there is no need to utilize a polarizer in the processing of the outputs. As with the embodiment of
One specific embodiment of the double spinner optical vector analyzer is shown in
Two of the polarization interactions are between different arms of the same spinner interferometer. These interactions are nulled using the polarization controllers in the interferometers (shown as LeFevre Loops 60 to distinguish from the nulling polarization controller 52 although they may all be LeFevre Loops). There are then four interactions remaining, and the four interactions are measurements of the four Jones matrix elements. Each matrix element is scaled by constant. These scaling values can be measured relatively easily, and, since they are determined by coupler splitting-ratios and splice losses, they should be stable over very long periods of time. The ability to calibrate each path loss out of the measurement should make it possible to obtain PDL measurements accurate to better than 0.05 dB.
A further advantage of this measurement method is the ability to use a single detector 64 for the measurement. As a result, gain matching between multiple detectors is not an issue, and relatively low precision (adjustable gain) components can be used. The use of a single device detector 64 also reduces the channel count required for a functioning system from four to three, meaning that a standard 4-channel board will now support a transmission and return-loss measurement.
Including all four elements in a single channel measurement does have the disadvantage of requiring more temporal range, and greater separation between the device, and any possible reflections (primarily connector reflections). In general, if a five nanosecond (5 ns) response is allowed (equivalent to 200 MHz sample spacing), then the fiber leads must be a minimum of 2 meters long. 3 meter leads would permit a 7 ns response time.
Scalability—Multilple Test Station Architecture
Because the double-Mach-Zender configuration of
A second application of the scalability of the double-Mach-Zender configuration would be to distribute the source light and the measurement light to multiple measurement stations for a plurality of devices under test (DUT 10) in a single facility. In this way the cost of the laser could be spread over multiple measurement stations, resulting in an extremely cost effective solution.
In view of the various embodiments and discussion of the present invention noted above, many variation and substitutions of elements for those disclosed will be obvious to one of ordinary skill in the art. Accordingly, the invention is limited only by the claims appended hereto.